AndreGuo / andreguo

Cheng Guo (郭铖/Andre) 👋

• 📕 Research interests:

HDR (High Dynamic Range), Inverse Tone-mapping, Tone-mapping, WCG (Wide Color Gamut), Gamut Mapping, IQA (Iamge Quality Assessment)

• 📫 Concact:

[e-mail], [e-mail], [e-mail]

• 🔭 Education:

2020-2024: Ph.D. at State Key Laboratory of Media Convergence and Communication (MCC), Communication University of China (CUC), Beijing, China

2022-2024: Visiting student at Deprtment of Media and Interaction, Peng Cheng Laboratory (PCL), Shenzhen, China

• 🌱 Current occupation:

2024-now: Dispaly algorithm engeneer (mini-LED, image quality) at TCL electronics, TV manufacturer.

Main Projects

1. ITM, inverse tone-mapping (SDR image to HDR/WCG):

🔭 CVPR2023

(Project: HDRTVDM; Model: LSN; Dataset: HDRTV4K):

Learning a Practical SDR-to-HDRTV Up-conversion using New Dataset and Degradation Models

• A luminance segmented network (LSN) AI model with channel decoupled self-attention, for inverse tone-mapping.
• New HDRTV4K training set and test set (SDR-HDR/WCG image pairs).
• New subjective metrics and objective assessment method on inverse tone-mapped HDR/WCG content.

🔭 CVMP2023:

(Model/Algorithm: ITM-LUT, plus an overview of AI-3D-LUT algorithms):

Redistributing the Precision and Content in 3D-LUT-based Inverse Tone-mapping for HDR/WCG Display

An efficient AI inverse tone-mapping for edge devices:

• AI learning of look-up table (LUT) content, and self-adaptability (LUT content will alter with input image) by the AI merging of basic LUTs.
• Run with fewer LUT size on higher-bit-depth (10/12bit) HDR/WCG, by discriminative non-uniform sampling of 3 smaller LUTs.

2. SI-HDR, single-image HDR reconstruction (SDR to HDR luminance)

🔭 ACCV2022:

(Model/Algorithm: LHDR):

LHDR: HDR Reconstruction for Legacy Content Using a Lightweight DNN

• An AI model for single-image HDR reconstruction, with partial convolution and condition.
• Lightweight design using mixed precision of network parameters etc.

3. TM, tone-mapping (HDR luminance to commom SDR image)

🔭 IEEE Access 2021:

Deep Tone-Mapping Operator Using Image Quality Assessment Inspired Semi-Supervised Learning

(Model/Algorithm: IQATM):

• An AI model for tone-mapping, with Laplacian Pyramid decomposition.
• Introducing IQA (image quality assessment) concept and metrics to unsupervised and semi-supervised training.